A total of 116 patients were included in this multicentre, randomized-controlled trial which set out to compare the efficacy of two antibiotic treatments of SBP. Trial was carried out in four hospitals in the north-east of Italy and was approved by the ethical committee of each hospital. Patients gave written informed consent to their participation. Inclusion criteria were: (i) diagnosis of cirrhosis with ascites and (ii) diagnosis of SBP. The diagnosis of SBP was based on a polymorphonuclear (PMN) count in ascitic fluid of >250/mm3 in the absence of clinical and radiological findings suggestive of secondary peritonitis.
Exclusion criteria were: (i) antibiotic treatment, including prophylactic treatment with quinolones, within 1 month of inclusion; (ii) history of hypersensitivity to quinolones or β-lactam antibiotics; (iii) age <18 and >75 years; (iv) evidence of other bacterial or fungal infections; (v) evidence of organic nephropathy (proteinuria, haematuria or abnormal findings on renal ultrasonography) and (vi) presence of shock, gastrointestinal (GI) bleeding, dehydration, hepatocellular carcinoma, cardiac failure, extrahepatic neoplasia at the diagnosis of SBP.
A total of 165 consecutive patients with cirrhosis and ascites with SBP were evaluated. Of these, 49 patients were excluded from the study for the following reasons: hepatocellular carcinoma (16 patients), treatment with antibiotics at diagnosis of SBP (nine), organic nephropathy (five), GI bleeding (five), age >75 years (three), hypersensitivity to β-lactam antibiotics (two), cardiac failure (two), extrahepatic neoplasia (two), other bacterial infections (two) and cardiac failure (three). Of 116 patients who were included in the study, 55 were randomly assigned to intravenous ceftazidime while 61 were assigned to switch therapy with ciprofloxacin.
Physical examination, chest and abdominal radiography, abdominal ultrasonography and routine laboratory tests were performed in all patients before starting the antibiotic therapy. In addition, samples of at least 60 mL of ascitic fluid and 20 mL of blood were taken for culture and other laboratory examinations at the patients’ bedside. Fresh urine sediment and a urine culture were also performed. Ascitic fluid PMN cell count and culture were repeated 48 h and 7 days after starting treatment. At least 20 mL of ascitic fluid were injected into two blood culture bottles (10 mL each). Patients discharged before the end of antibiotic treatment underwent these diagnostic procedures as out-patients.
Randomization was performed with sealed envelopes containing the treatment options prepared with random numbers generated by the statistica 6.1 software (StatSoft, Inc. 1984–2004, Tulsa, OK, USA). Randomization was independent for each hospital.
Intravenous ceftazidime (Spectrum, Sigma-Tau Sp.A., Pomezia, Rome, Italy) was administered at doses of 2 g b.d., 1 g b.d. and 1 g at every 24 h for serum creatinine levels of <1.5 mg/dL, 1.5–2.5 mg/dL, and >2.5 mg/dL respectively. Intravenous ciprofloxacin (Ciproxin, Bayer Sp.A., Milan, Italy) was administered at doses of 200 mg b.d. and 200 mg at every 24 h for serum creatinine levels of 2.5 mg/dL and >2.5 mg/dL respectively. Intravenous ciprofloxacin was followed by oral ciprofloxacin if and when clinical signs of infection (fever, abdominal pain, ileus, hepatic encephalopathy) had disappeared, PMN count in the ascitic fluid and in blood had significantly decreased (more than 50% of the baseline value). Oral ciprofloxacin (Ciproxin, Bayer Sp.A. Milan, Italy) was administered at doses of 500 mg b.d. and 250 mg b.d. for serum creatinine levels of 2.5 mg/dL and >2.5 mg/dL respectively. After the randomization patients received ceftazidime or cipropfloxacin for 8 days, resolution of SBP was established when all clinical signs of the infection (fever, abdominal pain, ileus, hepatic encephalopathy) had disappeared, PMN count in the ascitic fluid had decreased to a value lower than 250/mm3, total and differential white blood cell count had normalized, and ascitic fluid cultures were negative. In patients who did not respond to ceftazidime or ciprofloxacin, the antibiotic treatment was modified according to the in vitro susceptibility of the isolated organism or was modified empirically in patients with negative blood, urine and ascitic fluid cultures.
Treatment failure was considered whenever any of the following phenomena occurred: (i) no positive therapeutic response; (ii) recurrence of infection caused by the same organism within 1 month of the end of the assigned antibiotic treatment; (iii) superinfection caused by organisms resistant to the assigned antibiotic during treatment or within 1 month of its end and (iv) severe side-effects (i.e. allergic reactions, diarrhoea, dizziness, seizures).
In patients without renal failure at inclusion, renal impairment was diagnosed when the blood serum urea or serum creatinine level increased by more than 50% of the pre-treatment value, to levels higher than 50 mg/dL and 1.5 g/dL respectively. In patients with pre-existing renal failure, an increase in the blood urea nitrogen or serum creatinine level of more than 50% from baseline was required for a diagnosis of renal impairment. Type 1 HRS was diagnosed in patients who developed progressive renal failure despite the resolution of the SBP and had shown a final serum creatinine level of 2.5 mg/dL without any response to diuretic withdrawal and volume expansion by means of 1.5 L of isotonic saline solution, with proteinuria <500 mg/day, and with a normal renal ultrasound examination.35 Patients with type 1 HRS were fitted with a central venous line, a short peripheral intravenous catheter and a urinary bladder catheter. Terlipressin (Glypressin, Ferring Sp.A., Milan, Italy) was then given by continuous intravenous infusion starting with a dose of 2 mg/day. If no significant reduction in serum creatinine (≥30%) was observed, the dose was increased every 2 days in a stepwise manner to 4, 6, 8 and 12 mg/day. In addition to terlipressin, albumin (Kedrion Sp.A., Barga, Lucca, Italy) was given at a dose of 20–40 g/day in all patients to maintain central venous pressure between 12 and 14 cm H2O. Terlipressin and albumin were administered until the reversal of type 1 HRS or for a maximum of 15 days. Decrease of serum creatinine to <1.5 mg/dL was considered a complete response, while a decrease ≥50% of the peak creatinine value to a final value >1.5 mg/dL was considered a partial response to treatment for type 1 HRS. Patients were not given diuretics during the treatment of type 1 HRS. Patients who showed a complete or partial response to the treatment received diuretics as far as they were tolerated. Those patients with type 1 HRS as well as patients with type 2 HRS developing large ascites during the period of the study were treated with large volume paracentesis and albumin according to the International Ascites Club's guidelines on the management of ascites in cirrhosis.36 All patients included in the protocol were followed until death, liver transplantation or until 3 months after inclusion.
According to the current literatures17–19 we hypothesized similar efficacy for both antibiotic treatment regimens for SBP. The end point chosen to calculate the sample size was therefore the possibility of discharging patients before the end of antibiotic treatment (early discharge), taking into account that (i) an out-patient parenteral antibiotic treatment is not widely available and (ii) a third-generation cephalosporin is not available in both intravenous and oral formulations. Therefore, assuming that the possibility of an early discharge may occur in 25% of patients in the ciprofloxacin group and in no patient in the ceftazidime group, a minimum of 50 patients per group was required, with a two-sided type I error rate of 5% and a type II error rate of 20%.
The costs of antibiotic treatment and of hospital stay were calculated for each patient taking into account: (i) the type of antibiotic treatment, (ii) the presence of complications (type 1 HRS, hepatic encephalopathy and GI bleeding) during the stay in the Hepatology or Gastroenterology Unit and (iii) the stay in intensive care unit.
The results were analysed by a central coordinating committee at the Department of Clinical and Experimental Medicine at the University of Padova. Results were presented as mean ± S.E. Comparisons between groups were performed using the chi-square test or Fischer's exact test for categorical data, and Student's t-test for continuous data. Univariate analysis was used to identify factors predicting 3-month survival in patients with type 1 HRS. Variables reaching statistical significance in univariate analysis were subsequently included in multivariate analysis, by stepwise logistic regression, in order to identify independent predictors of the end point. Then, the survival functions were plotted according to the median value of the independent predictor variables. Statistical analyses were performed by using the statistica 6.1 software (StatSoft, Inc.). The P-values of <0.05 were considered significant.